S t a f f a n Wa l d o

T h e S w e d i s h I n s t i t u t e f o r F o o d a n d A g r i c u l t u r a l E c o n o m i c s

S L I W O R K I N G P A P E R 2 0 0 6 : 1

Capacity and Efficiency in Swedish Pelagic Fisheries

2

Capacity and Efficiency in Swedish Pelagic Fisheries

Staffan Waldo*

Abstract The aim of this study is to estimate capacity utilization and efficiency for Swedish vessels fishing in the pelagic segment in the period 1995-2002. The Swedish pelagic fishery is the largest fishery in Sweden both in terms of total landings and the value of the landings. Capacity utilization is estimated to 74 % using Data Envelopment Analysis (DEA). The development of the fleet is towards larger vessels, which indicates that larger vessels are preferred in the fishery. In the efficiency analysis, large vessels were found to be more efficient than small vessels on average. Also, newer vessels were found to be more efficient than older ones.

• Affiliation: The Swedish Institute for Food and Agricultural Economics (SLI), P.O. Box 730, S-220 07

Lund, Sweden, staffan.waldo@sli.lu.se.

3

1. Introduction According to economic theory, the problem with over-fishing is caused by

the absence of well-defined property rights for the marine resources

(Gordon (1954)). Without the possibility to claim a property right for the

fish, investments in a larger fish stock will not be beneficial for the

individual fisherman, since possible returns will be shared with others

participating in the fishery. Rather, as long as the fishery is profitable, the

competition for the “free” resource gives incentives for investments in

larger vessels and for new fishermen to enter. This so-called “race for the

fish” results in diminishing fish stocks and a fleet with excess fishing

capacity.

To regulate the fishery, the access to fisheries may be limited by licences,

catches may be limited by total allowable catches (TAC) per year, and how

and where to fish may be regulated by restrictions in the number of fishing

days, mesh sizes, closed areas etc. Such regulations may not necessarily

change the driving forces that exist in an unregulated fishery, and despite

the implementation of such regulations, European fisheries show many

symptoms that are characteristic for an unregulated fishery. The most

obvious result is that many commercially important fish stocks are over

fished. This does not only have biological consequences, but also economic

and social. The European Commission writes in its Green Paper that “The

fisheries sector is characterised by economic fragility resulting from over

investment, rapidly rising costs and a shrinking resource base: this is

reflected in poor profitability and steadily declining employment.”

Although the fish stocks form the base for fisheries, the economic

performance of the fishing fleet is not only dependent on the status of the

fish stocks, but also on an efficient fleet. The aim of the structural policy

within EU’s Common Fisheries Policy (CFP) is to modernize the European

fishing fleets, and subsidies are given both to modernizing vessels and to the

decommissioning of old vessels. However, subsidies may increase the

fleets’ capacity, and another issue in focus of the CFP is the relationship

4

between the fishing capacity and the size of the fish stocks. Until 2002 the

fleet was regulated through Multi Annual Guidance Programs (MAGPs),

but since 2003 the size of the fleet is to a larger extent and issue for the

individual member states.

Swedish fisheries management is part of the CFP. The Swedish

management does not define strong property-rights to the resource, so

economic theory predicts that excess capacity might be a problem.

However, the empirical knowledge about the topic is limited. The aim of the

study is to estimate capacity utilization and efficiency in the Swedish

pelagic segment. Pelagic fishing is the largest fishing in Sweden both in

terms of weight and catch value. The vessels in the pelagic segment mainly

target herring, mackerel and industrial species. Efficiency and capacity are

estimated using Data Envelopment Analysis (DEA). DEA is well suited for

modelling a sector with multiple outputs, such as vessels targeting more

than one species.

The paper continues with an introduction to Swedish fisheries in chapter 2,

where details of the pelagic fishery are presented. Chapter 3 contains a

discussion of the DEA method and how it is used for estimating capacity

utilization and efficiency. Empirical models and data are presented in

chapter 4. Chapter 5 contains the result concerning capacity utilization and

chapter 6 contains the results concerning efficiency. The results are

summarized and discussed in chapter 7.

5

2. Swedish Fisheries The Swedish fishery sector employs approximately 2200 licensed

fishermen. The major part of Swedish fishing takes place in marine waters,

employing 2000 fishermen. The most important fishing areas are the Baltic

Sea, the North Sea, the North Atlantic, the Skagerrak and the Kattegatt. In

2002, the total catch of marine species was 284 773 tons. In table 2.1 the

catch of different marine species is presented.

Table 2.1 Swedish Fish Landings in Tons Species Tons Eel 531Salmon 264Cod 15 115Saithe 1 583Herring 62 586Mackerel 5 090Industrial species 167 393Norwegian lobster 1 008Shrimps 2 151Other 29 052 Total 284 773 Source: National Board of Fisheries Sweden

The major fisheries are for cod and for pelagic species such as herring,

mackerel and industrial species. Cod is primarily caught in the Baltic Sea,

while herring, mackerel and industrial species are caught in large volumes

both in the Baltic Sea and in other fishing areas. Herring and industrial

species have the largest total landings. The landings of herring had a value

of 198 million Swedish crowns (SEK) and industrial species 211 million

SEK. The cod fishery has the highest economic value, 256 million SEK in

2002. (National Board of Fisheries Sweden (2003a)).

6

2.1 Swedish Fisheries Management1

Since 1995 Swedish fisheries is managed as a part of EU’s common

fisheries policy (CFP). Within the CFP, the regulations concerning the size

of the fleet are managed in the structural policy. Until 2002 this was done

through four development programs, MAGP I – MAGP IV. Subsidies for

vessel decommissioning were given in these programs, but in the CFP

subsidies were also given for modernization and construction. MAGP IV

covered the period 1997-2002 and is the development program that has been

in force most of the time of the Swedish EU membership. In 2003 the CFP

was reformed in favour of a more long-term management of fish stocks, and

today there are no investment subsidies for construction and the subsidies

for modernization have been more restrictive.

To protect the fish stocks from over-fishing, there are regulations

concerning how fishing is allowed. One such regulation is the amount of

fish allowed to be caught during the year, the Total Allowable Catch (TAC).

Most, but not all, species have a TAC. The member states decide how to

divide the national TAC between fishermen within the country. Many

Swedish TACs are divided into weekly vessel-quotas. The purpose is to

prevent the fishermen from taking all the allowable catches at the beginning

of the year. The weekly quotas are based on the vessels’ size. Swedish

fishing is also regulated by technical restrictions such as closed fishing

areas, temporary closing of specific fisheries and minimum mesh sizes. The

access to the fishery is restricted and it is not possible to become a

fisherman without a license from the National Board of Fisheries Sweden.

2.2 Segments

Swedish vessels are classified into five segments depending on the type of

fishery. The segments are:

• The coastal segment. The vessels have passive gear and are less than

12 meters. In 2002 there were 309 active vessels.

1 The section is based on National Board of Fisheries Sweden (2003b).

7

• The shrimp segment. The vessels are used for shrimp trawling. In

2002 there were 56 active vessels.

• The pelagic segment. The vessels are used for targeting herring,

mackerel and industrial species. In 2002 there were 124 active

vessels.

• Demersal segment. The vessels are used for targeting cod and

Norwegian lobster. In 2002 there were 180 active vessels.

• Vessels over 12 meters with passive gear. In 2002 there were 48

active vessels.

For the purpose of economic analyses, the segments are further divided

(Economic Assessment of European fisheries (2002)). The pelagic segment

is divided into vessels that are less than 24 meters and vessels that are 24

meters and over. The focus of the remainder of the paper will be on the

pelagic segment with vessels 24 m and over.

2.3 The Pelagic Segment, 24 m and over

The pelagic segment is the most significant Swedish fleet segment in terms

of capacity, volume and value of landings.2 Primarily pelagic and industrial

species like herring, sprat, mackerel, sand eel and blue whiting are targeted.

The segment is defined as vessels that are at least 24 meters. Vessels in the

segment primarily use trawling, but also purse seines.3 In 2002 Sweden had

58 active such vessels. The aggregate capacity was 21.9 thousand gross

tonnes (GT), or defined as engine power, 65.6 thousand kW. Employment

on board amounted to about 350 people. The total landings in 2002 were

247.8 thousand tonnes. Fishing takes place in the Baltic Sea, the North Sea,

the North Atlantic, the Skagerrack and the Kattegatt.

2.3.1 Fleet Development over Time The pelagic fleet has changed during the period 1995-2002 both in the fleet

structure and in catches and catch composition. The total tonnage has

increased by approximately 20%, at the same time as mean tonnage per

2 The section is based on National Board of Fisheries Sweden (2005). 3 A purse seiner puts a net in a circle around a school of fish and closes the lower end of the net to capture the entire school.

8

vessel has increased by about 40% from 271 gross tonnes to 393 gross

tonnes.

During the period 1995-2002 total available quotas decreased by

approximately 25% (available quotas are in general caught). Some

reallocation between species has occurred where e.g. traditional fishing in

the Baltic Sea and the Skagerrak/Kattegatt has decreased, while new fishing

for sand eel, capelin and North Atlantic herring have been introduced.

Figure 2.1 shows the development for the three fish categories analysed in

this paper, herring/sprat, industrial species and other species.

Figure 2.1 Catch composition

0

50000

100000

150000

200000

250000

300000

350000

1995 1996 1997 1998 1999 2000 2001 2002

Year

Tonn

es

Herring/sprat Industrial Other

As can bee seen from figure 2.1, herring and sprat are the dominating

species in Swedish pelagic fisheries. However, catches have declined

rapidly since 1998 while at the same time catches of industrial species have

increased. Other species are caught in very small quantities, although they

are important for individual vessels.

During 1995-2002 the number of vessels decreased from 65 to 58 while the

capacity in terms of both total gross tonnage and engine power increased. In

figure 2.2 the development of the number of vessels and total tonnage is

shown as an index with the year of 1995 having a value of 1.

9

Total tonnage increased in 1998 and has since been stable at about 20% over

the 1995 level. The number of vessels has decreased continuously, except

for 1998. In 2002 the total gross tonnage had increased by 18% compared to

1995 and the number of vessels had decreased by approximately the same.

2.3.2 Management of the pelagic segment

The management of quotas is important for over capacity since ill-defined

property-rights to the quotas will cause over investments in the fishery.

Dividing the TAC into vessel specific quotas is a system with stronger

property-rights than a system with free fishing until the entire TAC is

caught. The management of quotas differs in this respect between the target

species. The fishing for e.g. sand eel is free until the total Swedish TAC is

caught, while the majority of the herring and sprat quotas are formally

divided into two weeks quotas for each vessel depending on the size of the

vessel since 2002. The two weeks quota must be caught within the period

and can thus not be saved for future fishing. Before 2002 the fisheries

organization rationed the quota in a similar way. A few smaller quotas for

herring and the quota for mackerel are vessel specific, but not transferable.

Figure 2.2 Fleet development

0 0,2 0,4 0,6 0,8

1 1,2 1,4

1995 1996 1997 1998 1999 2000 2001 2002

Inde

x

Tonnage No. vessels

10

3. Data Envelopment Analysis (DEA) Efficiency and capacity are estimated using Data Envelopment Analysis

(DEA). In DEA, each production unit is compared with the best performing

units in a linear programming problem. Applied on fisheries, the idea is to

estimate the share of potential catches that a vessel has actually caught,

where potential catches are the catches observed for the best performing

vessels having the same (or less) inputs (e.g. tonnage, engine power, etc).

Efficiency and details of DEA modelling are discussed in section 3.1, and

capacity together with the relation between efficiency and capacity in

section 3.2.

3.1 Efficiency

DEA is a linear programming technique that is used to compare a vessel’s

in- and outputs with a best practice front. The best practice front constitutes

the maximum obtainable output for a vessel with a fixed amount of inputs

(this is called an output oriented approach). In the paper a vessel’s efficiency

index is defined as the vessel’s share of obtainable output.4

In DEA, the best practice front is spanned by the observations having most

outputs using a fixed amount of inputs. We do not know whether these best

performing vessels also are as efficient as they could be, i.e. if they are

efficient compared to a “true” front. It is possible that they too could

increase catches, although we have not observed any vessel having higher

catches given the observed input quantities. Efficiency in relation to a best

practice front and to a true front is illustrated in Figure 3.1.5

4 This is the reciprocal of the output oriented efficiency score defined in e.g. Färe et al

(1994). 5 Figure 1 illustrates what is called an output-oriented measure, where efficiency is estimated as the maximum possible expansion of outputs. Alternatively, efficiency may be estimated using input orientation, i.e. the maximum possible decrease in inputs. Which approach to use depends on what is the main objective of the fisherman, to increase outputs or decrease inputs. In the empirical application output orientation is used since it is assumed that the main objective of the fishermen is to maximize catches using available resources.

11

Figure 3.1. Efficiency In figure 3.1 four vessels are shown, a, b, c, and d. All vessels are assumed

to have the same inputs (i.e. identical vessels fishing the same number of

days). Each vessel produces two outputs (y1 and y2), e.g. two different target

species. Vessels a, b, and c are defined as efficient since none of the other

vessels is observed to catch more of both species. These vessels span the

best practice front, which is defined as the observed vessels and the straight

lines connecting them (linear combinations) as illustrated in the figure.

Vessel d is inefficient since it would be possible for it to catch more of both

species. The efficiency index is defined as the ratio Od/Od’. This ratio is

always less or equal to one, where efficient vessels obtain a value of one.

Although vessels a, b, and c are defined as efficient in a DEA estimate (they

cannot expand outputs and thus have an efficiency score of one), they are

not on the true frontier. If the true production possibilities were known these

vessels would also be defined as inefficient. However, the true front is not

known, and thus efficiency estimated with DEA is used as an estimate of

true efficiency.

The example in figure 3.1 illustrates a two-output case, but the idea may be

generalized to N inputs and M outputs. The efficiency index is estimated in

the following linear programming problem for a unit l:

d’

O

y 2 True front Estimated front a b d c 1 y

12

,

1

ln1

1/ max

, 1,...,

, 1,...,

0, 1,...,

l

z

K

k km lmk

K

k knk

k

Effindex

z y y m M

z x x n N

z k K

λλ

λ=

=

=

≥ =

≤ =

≥ =

∑

∑ (1)

X are the N inputs, y are the M outputs and z is a vector of activity

variables. K are the production units. Following this notation, ykm is the

m:th output for production unit k, and xkn is the n:th input for production

unit k. Efficiency is estimated using constant returns to scale (CRS) in

equation 1. A variable returns to scale (VRS) front is estimated by including

the constraint that the z-variables sum to one in the linear programming

problem. References to DEA literature are Coelli, Rao and Battese (2001)

and Färe, Grosskopf and Lovell (1994).

The efficient vessels span the production front to which inefficient vessels

are compared. Each point to which an inefficient vessel is compared is thus

a combination of some specific efficient vessels. These vessels are called

peers to the inefficient vessel. It is possible to identify which vessels are

used as peers for each inefficient vessel. This information may be used to

compare inputs and outputs of the evaluated vessels and their peers, but is

also useful for further analysis of vessel characteristics such as organization

and management. The number of times an efficient vessel serves as peer for

other vessels provides interesting information, since it tells us something

about how important the efficient vessel is for the efficiency scores of

inefficient vessels. So e.g. some efficient vessels are so-called “self

evaluators”, i.e. no other vessel is using them as peer. Removing such a

vessel from the sample will not affect the efficiency scores of other vessels.

In figure 3.1, vessel a is a self evaluator since it is not used as peer by any of

the other vessels. Vessels b and c are used as peers by vessel d.

13

3.2 Capacity Utilization

The definition of capacity used here is the one adopted by FAO (1998),

where capacity is defined as “the maximum amount of fish over a period of

time that can be produced by a fishing fleet if fully utilised, given the

biomass and age structure of the fish stock and the present state of the

technology”. The capacity utilization is the share of the maximum potential

catches that is actually caught.

The linear programming approach for estimating capacity is similar to that

for estimating efficiency. The difference is that when estimating capacity,

the variable factors of production are allowed to vary freely. The production

frontier in the capacity estimation is spanned by vessels that may have used

more variable inputs than the vessel under evaluation. As a consequence of

this, the difference between a vessel’s observed and obtainable output is not

only due to inefficiency, but also to the fact that the evaluated vessel has

used less variable inputs. Capacity utilization is estimated in a DEA model

as the ‘efficiency’ of a unit in relation to the production frontier where

variable inputs are unrestricted. In this paper, this is referred to as CU1. If

the fixed factors of production are xf and the variable factors are xv, the

linear programming problem for estimating CU1 for vessel l is

1 ,

1

ln1

1

1/ max

, 1,...,

, 1,...,

0, 1,...,

1, 1,...,

l

z

K

k km lmkK

fk kn

k

kK

kk

CU

z y y m M

z x x n N

z k K

z k K

λλ

λ=

=

=

=

≥ =

≤ =

≥ =

= =

∑

∑

∑

(2)

14

In equation 2 the variable factors of production are left out, since they do

not constrain capacity output.6 By imposing the restriction that the z-

variables sum to one, the model is estimated with a variable returns to scale

technology.

Färe et al (1989) argue that the CU1 measure may be downward biased since

outputs may be produced inefficiently. They propose a measure where a

technically efficient production is compared to the production frontier. This

measure, CU2, is calculated as CU1/Efficiency-index. The relation between

CU1, CU2 and the efficiency index is illustrated in figure 3.2 (following Färe

et al (1989))

Figure 3.2 Capacity and Efficiency

In the figure the obtainable output for vessels with the same amount of fixed

inputs is shown. By increasing the variable inputs, the obtainable output is

increased up to the f3-level, which is the largest observed catch. F3 is the

obtainable catch when estimating capacity utilization. Assume that an

observed vessel, a, has an output of f1. CU1 is then the ratio f1/f3, i.e. the

ratio between observed and obtainable catches. The efficiency index is

estimated as f1/f2, i.e. the ratio between observed catches and the catches

that are obtainable without increasing the variable inputs. CU2 is estimated

as f2/f3.

6 The variable inputs may be included in the linear programming problem to estimate optimal usage, see e.g. Färe et al (1989).

OutputsFå

f 3 CU 2

CU 1 f 2

Eff f 1 Variable inputs

15

4. Empirical Models and Data

Modelling Fisheries

When estimating efficiency with DEA no a priori assumption about the

functional form of the relationship between inputs and outputs is necessary.

The flexibility of the production relationships and the multiple input and

multiple output technology are the advantages of DEA. The drawback is

that DEA is not based on statistical theory, but is deterministic. The lack of

an error term may be problematic if there is a large random component in

the production. In fisheries there is clearly a component of luck, especially

if production is measured per trip. However, in the long run, luck will even

out but differences in managerial skill will persist. The importance of luck

and skill is discussed e.g. by Álvarez et al (2003).

Modelling production in a fishery requires that a number of considerations

are made concerning the production process and the production possibilities

of the individual vessels. One issue is how to deal with the catch of multiple

species (see Álvarez (2001)). Two species may be considered as different

outputs when they are targeted e.g. on different trips or with different gears.

On the other hand some species may to a large extent be caught in the same

fishing effort. Such species are not separable as different outputs. So e.g. the

by-catch of herring is large when targeting sprat.

DEA is well suited for a multi species characterization of the production

process. Efficiency is always estimated in relation to a point on the

production frontier that has the same catch composition as the vessel under

evaluation. That is, if a vessel primarily targets herring, its efficiency will be

evaluated in relation to other vessels primarily targeting herring. The

drawback with modelling a multi species production process is that when

the number of different outputs becomes larger, efficiency will tend to

increase due to characteristics of the DEA model. In this paper three output

dimensions are used: Herring/sprat, industrial species and other species.

Herring and sprat are aggregated to one output since it is not possible to

16

target the two species separately. Industrial species, such as sand eel and

blue whiting, are species that are targeted for reduction purposes. They are

primarily targeted in the North Sea. Vessels in the pelagic segment to a

minor extent also target other species. Species that are not pelagic or

targeted for industrial purposes are defined as a separate output.

On the input side, the fixed inputs gross tonnage and weight are used to

represent the size of the vessel. Engine power is important e.g. for trawlers,

which use their ability to move fast when fishing. Size statistics for the

vessels are used as standard within the CFP. Electronic equipment is not

shown in the vessel statistics, but is a potentially important tool to identify

where the fish is located. It will decrease search time and may thus increase

estimated efficiency if it is not accounted for as an input. Also, we do not

have access to variable inputs such as labour and fuel. The variable inputs

will vary with the number of fishing days and thus we use days at sea as a

proxy for them. Thus, in this paper three inputs are used: Tonnage, engine

power and days at sea.

As is the case in other industries that are based on renewable resources,

fishing is dependent on the size of the stock. When fish is abundant, catches

will be larger given the same effort. If not taken into account, fish

abundance will be included as a part of the estimated inefficiency. Including

stock estimates in the efficiency models is important when vessels face

different fish abundance. Many studies in fisheries include observations

from different periods of time, e.g. each month of the year. Here the fish

stocks are important because stock abundance changes between periods7 e.g.

by seasonal behaviour. In the case of Swedish pelagic fisheries, we estimate

efficiency separately for each year. Thus, we do not compare observations

from different periods of time. In this sense, all observations have the same

stock. What is more problematic is the heterogeneity of the fleet. Some

vessels only target herring and sprat, while others catch a large share of

industrial species. Vessels are also used for targeting the same species in

7 One reason for including different periods is to increase the number of observations when the number of vessels is small.

17

different fishing areas. This implies that the vessels face different stocks and

that the stocks are difficult to compare. One way of getting around this

problem is to use only a sub sample consisting of more homogeneous

vessels. This approach is used by e.g. Kirkley et al (1995), at the expense of

having only few vessels included. In the present study the entire Swedish

pelagic fishery is in focus and therefore all vessels are included.

Data

All data is administrated by the National Board of Fisheries Sweden.

Catches are estimated on board the vessel (reported in the vessel’s logbook)

and aggregated to total catches per vessel and year in the empirical

application. Data on tonnage, engine power and fishing days are used for

management purposes in the CFP. Data is available since the EU

membership in 1995 and the period studied is from 1995 to 2002. Data is

available for all vessels in the segment except a few (in 2002 e.g. three

vessels were discarded). The statistics for the pelagic segment below will

therefore be based on vessels used in the empirical analysis.

Descriptive statistics for inputs and outputs are presented in table 4.1.

Table 4.1 Descriptive Statistics for Inputs and Outputs 1995-2002

INPUTS OUTPUTS Variable Fixed

Mean values Days at Sea

Engine Power Tonnage Herring/sprat Industrial Other

1995 163,2 850,3 271,0 4094,9 853, 9 89,2 1996 192,2 885,6 287,2 4740,4 399,4 109,3 1997 198,6 922,3 298,6 4694,1 293,9 63,8 1998 186,5 958,7 336,4 5060,8 1093,5 47,4 1999 216,9 989,8 351,3 4438,1 1044,9 41,5 2000 209,5 1094,3 376,1 4286,6 864,3 84,0 2001 228,8 1143,9 383,8 3554,3 1338,5 114,7 2002 200,8 1165,5 393,0 2870,8 1952,8 57,9

The vessel size is increasing over the years. Measured as tonnage the mean

vessel has increased from 271 tonnes to 393 tonnes. The catch of herring

and sprat has been declining while industrial species have increased on

average. Days at sea have increased from 163.2 in 1995 to 200.8 in 2002.

The number of vessels varies between 53 in 2001 and 65 in 1995. A vessel

may be part of the segment for a number of years, leave for some other

18

fishery and then return to the pelagic segment again. 96 vessels are used in

the estimations but only 25 vessels have been part of the segment for the

entire period.

Age Distribution of the Fleet The vessels in the segment differ in characteristics like age, size and

catches. A description of differences between vessels with different

characteristics may therefore give useful information about the structure of

the segment. Also, the relation between vessel characteristics and the

efficiency results will be analysed in section 6.

The vessels are classified into three age categories, vessels built before

1970, vessels built from 1970 to 1994 and vessels built 1995-2002. Statistics

are presented in table 4.2 for the 55 vessels used in the empirical analysis

for 2002.

Table 4.2. Summary Statistics for Age Groups

No. Vessels Mean tonnage Mean das Mean catch/year

(ton) 1995-2002 9 689 234 8 740 1970-1994 24 419 224 5 226 -1969 22 244 162 1 966

Nine vessels that were fishing in 2002 have been built after the Swedish EU

membership in 1995. These vessels are significantly larger than the vessels

from other age groups, and their mean catches are also higher. The oldest

vessels are the smallest on average, they fish fewer days and also have the

smallest mean catch. The vessels built between 1970 and 1994 are the

second largest on average and they fish approximately the same number of

days as the most recently built vessels. The age classification is based on the

construction year of the vessel and modernizations may have taken place

over the years.

The catch composition differs somewhat between the age groups. In table

4.3 the catch composition for 2002 is presented. All groups of vessels have

herring/sprat as the largest share of the catch and very small catches of

19

“other species”. Industrial species constitutes 43% of the catch for the

vessels built 1995-2002, which is the largest share for the three groups. For

the vessels built 1970-1994, the industrial species are 38% of the catch and

for vessels built before 1970 industrial species constitutes only 16% of the

total catches. Table 4.3 Catch Compositions for Age Groups 1995-2002 1970-1994 -1969 Herring/sprat 57% 62% 83% Industrial 43% 38% 16% Other 0% 1% 1%

Differences in catch composition between the age groups reflect the

geographical location of the vessels. The most recently built vessels are

located on the Swedish west coast, which is closer to the fishing areas where

industrial species are caught. A larger share of the older vessels is located

on the south coast, where mainly herring and sprat are targeted.

20

5. Capacity Utilization Capacity utilization has been estimated for each vessel for 2002. The mean

values are presented in table 5.1. Recall from figure 3.2 that CU1 is the

entire difference between actual catches and potential catches, while CU2 is

the difference net of estimated inefficiency.

Tabell 5.1 Capacity utilization

Mean CU1 0,743 CU2 0,878

The mean value for CU1 is 0.74, which implies that the catches of the

vessels were on average 74% of potential catches. 55 vessels were active in

the segment. Twelve of these utilized their full capacity. Among the vessels

with full capacity utilization vessels from both the south coast and the west

coast are represented, as well as both large and small vessels. 15 vessels had

capacity utilization (CU1) of less than 60%. Also among these there are both

large and small vessels represented and vessels both from the south and

west coasts of Sweden.

CU2 is 0.88. In the CU2 measure, the efficiency index is not part of the

capacity utilization. Instead, a low CU2 value depends only on the use of

variable inputs. Thus, capacity utilization estimated as CU2 is always larger

than if estimated as CU1.

The total capacity of the segment is the sum of the capacity of each vessel in

the segment. If all vessels were fully efficient and used their entire capacity,

they would be able to catch 325 100 tonnes, which can be compared with

the actual catches of 247 800 tonnes. The excess capacity is thus 77 300

tonnes, which is approximately 31% of actual catches or 24% of the

capacity.

The analysis shows that the catches in 2002 could have been made with

considerably fewer vessels, if these vessels hade used their full capacity.

21

One example of such a decrease in capacity could be if the vessels with the

highest efficiency index had stayed in the segment. To generate the catches

for 2002, it would have been sufficient for the segment to operate with 41

vessels. These correspond to approximately 75% of the number of vessels in

the segment in 2002.

Excess capacity is not limited to fisheries but is found in most industries.

The demand for products tends to vary during the business cycle and it may

be profitable for a company to have excess capacity to meet an increase in

demand. The fisheries sector, however, is different since the fish is a

common resource. The open access problem leads to excess capacity which

is too large.8 On the other hand it may be rational to have some excess

capacity in order to adjust fishing to biological fluctuations in the fish

stocks. It may also be rational to have high capacity in the short run if

actions are taken to increase the stocks, and thus more capacity may be

needed in the future.

However, the general problem is that too much fishing capacity is in use.

The European commission states in the Green Paper that “the available

fishing capacity of the Community fleets far exceeds that required to harvest

fish in a sustainable manner”.

International Comparisons

Comparing the results with studies from other fisheries must be interpreted

with caution since these are calculated under different resource conditions

and different regulatory systems. Lindebo, Hoff and Vestergaard (Lindebo

(2004, chapter 2)) have estimated the capacity utilization for Danish North

Sea trawlers in 1999. These are used for targeting cod, herring/mackerel and

industrial species. They find a CU1 value of 0.81 and a CU2 value of 0.94.

The capacity utilization is thus somewhat higher in this case compared to

Swedish pelagic fisheries. Lindebo (2004, chapter 4) analyses the flatfish

fishery in the North Sea for vessels from Denmark, France, the Netherlands,

8 See e.g. Brady (2004).

22

and the United Kingdom. Capacity utilization is estimated to be 77% for the

entire fishery, but it varies substantially between the countries. Vestergaard

et al (2003) estimate capacity utilization for Danish gill-netters in the North

Sea and the Skagerrak. They find CU1 to be 0.88 and CU2 to be 0.92.

Dupont, Grafton, Kirkley and Squires (2002) estimate a capacity model

where it is possible for the vessels to make non-radial increases in the

catches. The fishery studied takes place in Canada with active gears for e.g.

cod and haddock. Capacity utilization varies from 0.64 in 1990 to 0.72 in

1998. Compared to these studies, the capacity utilization in the Swedish

pelagic segment is approximately the same as for other fleets, although the

Danish studies show somewhat higher capacity utilization for both the CU1

and CU2 measures.

Studies not using the DEA approach are Bjørndal and Gordon (2000) who

only find a small over capacity in Norwegian pelagic fisheries in 1994-96

and Nøstbakken (2005) who find considerable over capacity in the same

segment for the period 1998-2000. Eggert and Tveterås (2004) studies the

Swedish trawl fishery for cod in the Baltic Sea. They estimate over capacity,

given the small fish stocks, to approximately 25 %. They also find

economics of scale in the segment and that a structural adjustment of the

fleet could cause cost savings of about 40%. The over capacity for Swedish

cod trawlers is about the same as is estimated in the CU1 measure for the

pelagic segment.

23

6. Efficiency Results In section 6 the efficiency scores are presented for efficiency estimated both

under VRS (6.1) and CRS (6.2) assumptions. In sections 6.3 to 6.5 an

analysis of efficiency differences between vessels is performed. The

analysis deals with size differences, age differences and vessel

decommissioning.

6.1 Variable Returns to Scale

Estimated mean efficiency ranges from 0.80 in 1995 to 0.89 in 2001. In

table 6.1 the efficiency scores are presented for the entire period 1995-2002. Table 6.1 Summary Statistics for Efficiency Scores (VRS) Min Max Mean Std. dev No. vessels 1995 0,36 1,00 0,80 0,19 65 1996 0,25 1,00 0,81 0,21 57 1997 0,45 1,00 0,82 0,20 58 1998 0,37 1,00 0,81 0,22 64 1999 0,30 1,00 0,83 0,21 56 2000 0,36 1,00 0,87 0,18 59 2001 0,39 1,00 0,89 0,16 53 2002 0,50 1,00 0,84 0,16 55

The interpretation of the result for 2002, where mean efficiency is 0.84, is

that the vessels on average caught 84% of potential catches.9

Examining the individual vessels further, we find that in 2002 the efficiency

ranges from 0.5 to 1. In figure 6.1 the vessels are arranged according to the

estimated efficiency. The least efficient vessel is located to the left in the

figure and the fully efficient vessels to the right.

9 Eggert (2001) estimates efficiency for Swedish Nephrops trawlers using a stochastic production function and data per fishing trip. Mean efficiency is 66% in his study.

24

17 vessels are defined as efficient. The other vessels are inefficient and

could thus increase catches without increasing resources. 12 vessels have an

efficiency score below 0.75.

A problem in DEA estimations is that the model does not take technological

differences into account. Thus, part of the efficiency index may be due to

the fact that the vessels have different technology. As shown in table 4.2,

there is a clear correlation between the age and the size of the vessels. It is

reasonable to think that newer vessels have more advanced technology than

older, although investments may have been made to adapt older vessels to

modern fishing. Since differences in scale are taken into account in the VRS

estimations, technological differences may also be taken into account to

some extent.

6.1.1 Peer vessels

The efficiency index is always estimated in relation to one or more peer

vessels. An analysis of the peer vessels may give useful information for

instance on the comparability between a vessel and its peers, or it may show

if a large number of the peer vessels have certain characteristics in common.

Here, the analysis is restricted to a comparison of inputs between peer

Figur 6.1 Efficiency (VRS) 2002

0 0,2 0,4 0,6 0,8

1 1,2

0 10 20 30 40 50

Effic

ienc

y

25

vessels and other vessels. Mean values for inputs are calculated for peer

vessels and other vessels separately. For each input a ratio is calculated as

the mean value of the peer vessels divided by the mean value of the other

vessels. So for tonnage e.g., a value larger than one implies that the peer

vessels are on average larger than the other vessels. The ratios for the inputs

days at sea (das), engine power (kwH) and tonnage (EU-ton) are presented

in figure 6.2

Figure 6.2 Differences between peer vessels and other vessels (VRS)

0

0,5

1

1,5

2

1995 1996 1997 1998 1999 2000 2001 2002

Rat

io

Das KwH EU-ton

The results show that the peer vessels are approximately the same size or

somewhat smaller than the other vessels, as the ratios are below or

approximately equal to one. The interpretation is that on average the vessels

are not compared with larger vessels in the estimations. This is an advantage

if the fishing possibilities (such as the length of fishing trips or sensibility to

weather conditions) are different for small and large vessels.

6.2 Constant Returns to Scale

Summary statistics for efficiency estimated using CRS is presented in table

6.2 Table 6.2 Summary Statistics for Efficiency Scores (CRS) Min Max Mean Std. dev. No. vessels 1995 0,29 1,00 0,72 0,21 65 1996 0,25 1,00 0,69 0,24 57 1997 0,21 1,00 0,68 0,25 58 1998 0,17 1,00 0,63 0,26 64 1999 0,27 1,00 0,75 0,23 56 2000 0,20 1,00 0,75 0,24 59 2001 0,35 1,00 0,80 0,19 53 2002 0,21 1,00 0,72 0,21 55

26

Mean efficiency for 2002 is 0.72. The interpretation is that vessels on

average catch 72% of potential catches. Observe that the mean efficiency

scores are lower for the CRS estimates than for the VRS estimates in each

of the years. Mean efficiency shows a development similar to that in the

VRS case, although the differences between the years are larger. The

distinct decline in 1998 is of special interest. It does not occur in the VRS

estimates, although the mean efficiency score is the smallest also for the

VRS case in 1998. We also note that to a large extent efficiency follows the

development of herring and sprat catches with a peak in 1998 followed by a

decline in the coming years. In 1998 Sweden also gained access to a sand

eel quota in the North Sea.

6.2.1 Peer Vessels The relations of the input resources for peer vessels and other vessels are

presented in figure 6.3.

As can be seen in figure 6.3, the peer vessels were larger all the years, but in

2001 the engine power of the peer vessels was less than that of the others

(the ratio is less than one). The general picture though is that peer vessels

are larger on average.

In 1998, when CRS efficiency had a peak, the ratios for all inputs are high.

With the exception of 2002, the differences between peers and other vessels

are at their largest this year. This might be explained by the fact that larger

vessels were able to better utilize the increased fishing possibilities. When

Figure 6.3 Differences between peer vessels and other vessels (CRS)

0

0,5

1

1,5

2

1995 1996 1997 1998 1999 2000 2001 2002

Rat

io

das KwH EU-ton

27

estimating efficiency assuming CRS, the scale of operation is not taken into

account. Thus, small vessels may be compared to a point on the frontier that

is a down-scaling of the production of a large vessel. Low efficiency may in

the CRS case stem from an inefficient scale of operation. The relation

between vessel size and efficiency is analyzed in section 6.3.

6.3 Size and Efficiency

The peer vessels are on average larger than other vessels when the

efficiency index is estimated by using constant returns to scale. This

indicates that it is an advantage to use larger vessels. An interesting question

is therefore if large vessels systematically have higher efficiency indices

than smaller vessels. If vessels of a certain size have higher efficiency

indices than others, there could be a reason for analysing if they are better

suited for pelagic fishing.

The vessels are divided into four size categories: Vessels less than 200

tonnes, vessels 200-399 tonnes, vessels 400-599 tonnes and vessels larger

than 600 tonnes. The development from smaller to larger vessels during the

period 1995-2002 is clear from figure 6.4.

Figure 6.4. Number of Vessels in Size Classes

05

10152025303540

1995 1996 1997 1998 1999 2000 2001 2002Year

No.

Ves

sels

-199 200-399 400-599 600-

The number of vessels in the two smallest size classes has decreased during

the period, while the number of vessels in the two largest size classes has

28

increased. In 1995, 20 vessels under 200 tonnes operated in the segment,

whereas in 2002 only 7 such vessels were left. The number of vessels

between 200 and 399 tonnes decreased from 34 to 23, while the number of

vessels between 400 and 599 tonnes increased from 8 to 19. Only few

vessels are larger than 600 tonnes, 3 in 1995 and 6 in 2002.

Estimated mean efficiency using a CRS technology for the four size classes

is presented for the period 1995-2002 in figure 6.5.

It is clear from the figure that the two smallest classes have lower efficiency

indices than the two largest. The size differences indicate that larger vessels

are preferred from an efficiency perspective. When interpreting the results it

is important to note that the development of the sector towards larger

vessels implies that larger vessels will tend to be newer. Thus, the high

efficiency scores for small vessels may to some extent be due to old fishing

techniques that have not been taken into account in the efficiency model.

6.4 Age and Efficiency

Vessels built between 1995 and 2002 can be expected to function as peer

vessels in the analysis because it has been possible to adapt the size and

engine power of these vessels in accordance with the fishing opportunities

that existed in the segment during the studied period. The vessels are e.g.

larger than older vessels as discussed in section 4. Newer vessels may also

Figure 6.5 Efficiency (CRS) for size classes

0 0,2 0,4 0,6 0,8

1 1,2

1995 1996 1997 1998 1999 2000 2001 2002

Mea

n ef

ficie

ncy

- 199 200-399 400-599 600-

29

to a larger extent fit modern fishing techniques. 11 vessels were built during

the period 1995-2002 (9 vessels were active in the segment in 2002). In this

section for the sake of simplicity we classify vessels built in 1995-2002 as

modern, although of course other vessels also may have modern equipment.

This, however, cannot be deduced from available data. The hypothesis is

that modern vessels have higher efficiency indices than older vessels and

that the modern vessels define the technological front with which others are

compared.

The first year analysed is 1998. Three modern vessels were built before

1998 and two during that year. The constant returns to scale model is used

since the question studied is caused by long-term decisions made by the

fishermen. When investing in new vessels, decisions concerning size,

engine power, etc, are choice variables and investments that are not on an

optimal scale are defined as inefficiency in the CRS model. The same

models as in section 6.2 are used, but the results are divided into vessels

built before 1995 and vessels built in 1995-2002.

Results from the efficiency analysis are presented in table 6.3. The third

column shows the total number of vessels in each age category. The fourth

column shows the mean efficiency of these vessels. The number of vessels

defined as efficient and how often these are used as peers for others is

presented in the two last columns.

30

Table 6.3 Efficiency for Vessels Built Before 1995 and Vessels Built 1995-2002 Year Age

Category No. Vessels Mean

efficiency No.

Efficient Serves as

Peers

1998 1995-2002 5 0,73 1 2

-1995 59 0,62 10 131

1999 1995-2002 5 0,93 3 26

-1995 51 0,73 8 121

2000 1995-2002 8 0,75 1 14

-1995 51 0,75 14 118

2001 1995-2002 8 0,87 1 16

-1995 45 0,79 10 108

2002 1995-2002 9 0,93 3 37

-1995 46 0,68 5 87

The first observation is that the vessels built in 1995-2002 on average have

higher efficiency indices in each of the studied years, except for 2000.10

The number of peer vessels, however, is comparatively low for all the years,

and the peer vessels are not used as peers for many other vessels either. So

e.g. only one modern vessel functioned as a peer vessel in 1998, and this

only twice. The other ten peer vessels in 1998 were older vessels. The

interpretation is that many of the older vessels also have large catches in

relation to the resources used. During the entire period 1995-2002, the

modern vessels are high-performing but do not constitute a predominant

part of the segment in terms of being peers to others. A possible explanation

is that the modern vessels to some extent are used for targeting other species

than older vessels.

6.5 Were Scrapped Vessels Less Efficient?

A crucial part of the CFP is to decrease the size of the fleet in order to

achieve a balance between fish stocks and fishing capacity. Subsidies are

granted for the decommissioning of vessels. It may be assumed that it is

predominantly vessels with low efficiency scores that are decommissioned

10 The result is in line with Andersen (2002b) who finds older Danish trawlers to be less efficient than newer.

31

(Vestergaard et al (2003)), since these vessels achieve lower catches with

the same use of resources.

Data is available for vessels that have been decommissioned with financial

support from the CFP. During the period 1995-2002, there were four such

vessels. Data is available for the last year these vessels were active in the

pelagic segment, but the vessels may have been active in other segments

before being decommissioned.

The efficiency analysis shows that two of the decommissioned vessels had

an efficiency index that was distinctly lower than average for both constant

and variable returns to scale. The other two had average performance, which

indicates that it is not necessarily only low-performing vessels that are

decommissioned with subsidies from the CFP. However, with only four

decommissioned vessels during the entire period it is not possible to draw

any general conclusions.

32

7. Summary and Discussion The importance of well-defined property rights in fisheries is strongly

emphasized in economic theory. In an unregulated fishery, the theory

predicts excess capacity, over-fishing, and low profitability. The symptoms

are well known, and fishing fleets around the world have long been

regulated by governments in order to deal with the problems. Examples of

this are the management plans for fishing capacity in the EU, and FAO’s

international plan of action for fishing capacity. The future of the fishing

sector is dependent on a balance between fleet capacity and biological

resources. However, the long term economic development is also dependent

on an efficient fleet where the fish is caught by using a minimum of

resources. In this paper capacity and efficiency are analysed for Swedish

vessels fishing in the pelagic segment between 1995 and 2002. The pelagic

segment is the most important segment in Swedish fisheries both in terms of

volume and value of landings. Vessels in the pelagic segment are primarily

used for targeting herring, sprat, mackerel, and industrial species.

The present management of the pelagic fleet does not affect the basic

features concerning property rights in the fishery. The pelagic species are

regulated by free fishing under the TAC, two-week rations, and vessel

specific quotas (these are only smaller quotas). A majority of Swedish

fishermen are allowed to fish for pelagic species, so there is a potential for

an increase in the number of vessels if the segment becomes more

profitable.

The results of the empirical analysis show that the average capacity

utilisation was 74 % in 2002. Both large and small vessels as well as vessels

from the south and west coasts of Sweden are represented among vessels

with both high and low capacity utilization. Calculating capacity utilisation

as the number of vessels, it can be shown that the 41 vessels with highest

technical efficiency would be able to generate the observed catches for 2002

33

if they were to utilize their capacity fully. 55 vessels were part of the pelagic

segment in 2002.

Excess capacity is not a phenomenon that is unique to the fishing industry.

In most industries the demand varies over time, and having excess capacity

will enable a firm to increase its production when demand is high. In

fisheries the stocks and thereby the fishing opportunities may vary

considerably due to biological conditions. Some excess capacity may thus

be profitable in the long run so that the fisherman can take advantage of

periods with good fishing. However, the fishing sector is different from

many other sectors because of the common ownership of the fish resources,

and economic theory predicts an excess capacity that is not motivated by

changes in fishing possibilities.

The efficiency of the vessels is studied in the paper with regard to the trend

in the segment that newer vessels are larger than older. Investments have

been made in large vessels, which indicate that larger vessels are better

suited for pelagic fishing. The empirical results show that larger vessels in

general have higher efficiency than smaller ones in the period studied. This

confirms the hypothesis that there is economics of scale in the segment. The

size difference between old and new vessels is especially manifest for

vessels built before and after Sweden became a member of the EU in 1995.

The newer vessels are considerably larger than the older and are expected to

be more efficient since they are built to utilize the fishing possibilities

optimally during the period studied. The analysis also shows that the

modern vessels are more efficient on average. Thus, the general trend is

towards larger vessels, and the high efficiency of these does not indicate any

changes in this trend. However, a change of the regulations may also change

the optimal size of a vessel.

34

References Álvarez A. 2001. Some Issues in the Estimation of Technical Efficiency in a Fishery. Efficiency Series Paper 02/2001, University of Oviedo. Álvarez A, Pérez L and Schmidt P. 2003. The Relative Importance of Luck and Technical Efficiency in a Fishery. Efficiency Series Paper 03/2003, University of Oviedo. Andersen Jesper L. 2002a. Using different inputs and outputs to estimate technical efficiency in fisheries. Danish Research Institute of Food Economics, Working Paper 10/2002. Andersen Jesper L. 2002b. Reasons for Technical Inefficiency of Danish Baltic Sea Trawlers. Danish Research Institute of Food Economics, Working Paper 18/2002. Bjørndal T och Gordon D. 2000. The economic structure of harvesting for three vessel types in the norwegian spring-spawning herring fishery. Marine Resource Economics 15. pp 281-292. Brady M. 2004. Fiske i framtiden – hur förvalta en gemensam naturresurs. Swedish Institute for Food and Agricultural Economics, report 2004:5. In Swedish, available at www.sli.lu.se. Coelli T, Rao P and Battese G. 2001. An introduction to efficiency and productivity analysis. Kluwer Academic Publishers. Commission of the European Communities. 2001. Green Paper on the Future of the Common Fisheries Policy. Eggert H och Tveterås R. 2004. Potential Rent and Overcapacity in the Swedish Baltic Sea Trawl Fishery. Working Papers in Economics No 152. Nationalekonomiska institutionen. Göteborgs universitet. Farrell M.1957. The measurement of productive efficiency. Journal of the Royal Statistical Society pp 253-81. FAO. 1999. Fisheries Technical Paper No 386, Rome 1999. Färe R, Grosskopf S and Lovell K. 1994. Production Frontiers. Cambridge University Press. Fødevareøkonomisk Institut FOI. 2002. Common Fisheries Policy reform – A new fleet capacity policy. Kirkley J, Squires D and Strand I. 1995. Assessing Technical Efficiency in Commercial Fisheries: The Mid-Atlantic Sea Scallop Fishery. American Journal of Agricultural Economics 77, pp 686-697. Kirkley J, Squires D and Strand I. 1998. Characterizing Managerial Skill and Technical Efficiency in a Fishery. Journal of Productivity Analysis 9, pp145-160. National Board of Fisheries Sweden. 2001. Betydelsen av de kvoter som står till det svenska yrkesfiskets förfogande. National Board of Fisheries Sweden. 2003a. Fakta om svenskt fiske och fiskkonsumtion.

35

National Board of Fisheries Sweden. 2003b. Utveckling av regleringsformer. Promemoria 102-147-03. National Board of Fisheries Sweden. 2005. The Development of a Swedish Pelagic Segment in the context of EU Structural Support Schemes 1995-2002. Nøstbakken L. 2005. Cost Structure and Capacity in the Norwegian Pelagic Fisheries. Presenterad på European Association of Fisheries Economists konferens nr XVII i Thessaloniki, Grekland. Pascoe S, Hassaszahed P, Andersen J and Korsbrekk K. 2003. Economic versus physical input measures in the analysis of technical efficiency in fisheries. Applied Economics 35. pp 1699-1710. Squires D, Grafton Q, Alam, M.F and Omar I.H. 2003. Technical efficiency in the Malaysian gill net artisanal fishery. Environment and Development Economics 8, pp 481-504. Vestergaard N, Squires D, Jensen F and Andersen L J. 2003. Technical Efficiency of the Danish Trawl Fleet: Are the Industrial Vessels Better Than Others?. Nationaløkonomisk Tidskrift 141, pp 225-242.